NDT,Normal Distributions Transform正太分布变换方法。
The normal-distributions transform (NDT) method for registration of 2D data was introduced by Biber and Straßer。该方法最早由Biber and Straßer提出。
The NDT models the distribution of all reconstructed 2D-Points of one laser scan by a collection of local normal distributions. First, the 2D space around the robot is subdivided regularly into cells with constant size. Then for each cell, that contains at least three points, the following is done:
1) Collect all 2D-Points xi=1..n contained in this box.
2) Calculate the mean q.
3) Calculate the covariance matrix Σ.
The probability of measuring a sample at 2D-point x contained in this cell is now modeled by the normal distribution N(q, Σ)。
标准的NDT方法
多尺度NDT
聚类NDT/区域生长NDT
NDT-MCL
参考文献
[1] Peter Biber and Wolfgang Straßer. The normal distributions transform:A new approach to laser scan matching. In Proceedings of the IEEE International Conference on Intelligent Robots and Systems (IROS), pages 2743–2748, Las Vegas, USA, October 2003.